New radiology research highlights inherent ‘unfairness’ in RVU system

New research highlights what experts believe is inherent “unfairness” in the system used to calculate radiologist payments. 

Work relative value units (RVUs) measure a physician’s clinical productivity including the period it takes to perform a service. In musculoskeletal imaging, this time correlates directly with the number of images produced per X-ray, researchers wrote Tuesday in Current Problems in Diagnostic Radiology. However, certain Current Procedural Terminology (CPT) billing code descriptions tied to RVUs leave ambiguity as to the number of image views involved in an exam. 

This can lead to radiologists being paid the same amount, regardless of whether referring clinicians ask them to examine two images or several extra as part of a procedure. For instance, “X-ray of a foot with at least three views” (CPT code 73630) could result in a radiologist reading three scans, five, or even more.

“[Centers for Medicare & Medicaid Services] rules and guidance documents explain that obtaining extra views for quality assurance or diagnostic clarification purposes should not be billed as an extra study,” radiologist Oganes Ashikyan, MD, with UT Southwestern Medical Center in Dallas, and co-authors wrote Jan. 14. “However, wRVUs for that specific CPT code remain identical and do not capture the workload of evaluating the extra images. Preferences of referring clinicians, experience of technologists and other factors vary among practices. As such, chances that one group of musculoskeletal radiologists is evaluating more images compared to their counterparts in another practice increase.”

To illustrate their point, Ashikyan and colleagues queried the organization’s database of over 76,000 MSK reports. They reviewed a sample of 440 random radiographs to evaluate variability in the number of images obtained. Their final sample consisted of 10 studies from each of the 44 musculoskeletal CPT codes. This included 242 studies from a safety net healthcare system and another 198 from university-associated hospitals and clinics. 

About 31% of studies (75/242) were found to have a “mismatched” number of images from the safety net sample (meaning one-third used more X-rays than specified in the CPT code description). That’s compared to 33% of exams (66/198) on the university health system side. Ashikyan and co-authors highlighted “significant” differences between the two data samples, with more studies using extra images in the safety net setting. Variability held true for both radiographs with CPT codes outlining a “minimum” number of required views and those with a precise figure. 

“The exact reasons for the differences between our two healthcare systems are difficult to determine,” the authors noted. “The potential explanations include differences in ordering patterns, preferences of clinicians, and differences in training and experience of technologists. Our university system has more clinics staffed by specific surgical faculty. Some surgical faculty [prefer] to include two extremities on one image.”

There are various potential solutions to make the workload metric more valid and reliable, UT Southwestern experts added. Some institutions have created a “local value system to more accurately reflect their radiologists’ workloads.” Others have devised “clever” ways to normalize data, such as multiplying an RVU by the number of studies. Previous research also has explored utilizing natural language processing tools to consider the complexities in radiology reports, with reading times estimated from such scores. 

“These solutions require resources as they are being created, evaluated, and implemented,” the authors cautioned. “In situations where CMS-assigned wRVUs [are] the only practical solution, using a range of percentiles rather than specific percentiles will remove some of the unfairness that results from variability in workloads. For example, instead of comparing radiologists to [a] specific percentile, broad ranges such as quartiles or quintiles would make the assessment less prone to inaccuracies and errors.”